49results about How to "Improve the accuracy of face recognition" patented technology

The invention provides a Maxout multi-convolution neural network fusion face recognition method. The technical problem of unequal classification of positive and negative samples in a single network is solved. The method comprises the steps that a regression tree combination algorithm and affine transformation processing training data are used to acquire a database after alignment; face images in the aligned database are extracted to acquire a database simply with face area images; two networks are built, and a Maxout method is used to acquire a fused network; the database simply with face area images is used to train the fused convolution nerve network to acquire a trained network model; and a test image is pre-processed to test the trained network model. According to the invention, a Maxout module is used to fuse two or more subsystems, and then is successively connected with a full connection module and an output module to acquire a complete face recognition system; the classification sensitivity of positive and negative samples of the convolution neural network can be balanced; and the face recognition accuracy is improved.

The invention provides a face characteristic extraction modeling method based on a neural network, and a face characteristic extraction modeling device. The face characteristic extraction modeling method comprises the steps of acquiring a triple face picture from a training set of a preset application scene; acquiring a trained face identification neural network, and determining a triple deep neural network according to the face identification neural network; using the triple face picture as input of the triple deep neural network, and determining a loss cost function value; training the triple deep neural network according to the loss cost function value and a training parameter; performing face identification testing on the triple deep neural network through the training set of the preset application scene, determining the testing precision according to the testing result, and determining an objective face feature extraction model according to the testing precision and the triple deep neural network. The face characteristic extraction modeling method and the face characteristic extraction modeling device can realize a beneficial effect of high face identification precision when a face feature model which is obtained through modeling is applied in a specific face identification application scene. The invention further provides a face identification method and a face identification device.

The invention discloses a fast face angle recognition method based on deep learning. Face angle recognition relates to rotation angles of a head around three axes in a three-dimensional vertical coordinate system, and a face angle change can cause partial absence of facial feature information. The method comprises the following steps: (1) labeling face images at times of rotation of 0 degree, plus and minus 15 degrees, plus and minus 30 degrees and plus and minus 45 degrees to establish a database; (2) using angle labels to carry out feature extraction through five layers of a convolution module, transforming features into 2048-dimensional feature vectors through two layers of full connection, using the feature vectors as inputs of a softmax classification layer, carrying out classification, and thus establishing a convolutional neural network structure; and (3) based on an ubuntu 16.04 operating system, training the convolutional neural network under a GPU of 1080 and a deep learning framework of a CAFFE to obtain a face angle recognition model. The method is used for deep learning-based real-time recognition of three-dimensional face angles.

The invention discloses a non-cooperative face recognition method based on depth learning, which comprises the following steps: 1, generating a depth learning network training data set; 2, respectively construct a detection depth learning network and a recognition depth learn network; 3, respectively train a detection depth learning network and a recognition depth learn network; 4, prepare a non-coordinated face feature database; 5, that camera sample the video stream in real time; 6, detect and tracking that face region of the image; 7, feature match; 8. Face Recognition. By introducing a tracking algorithm in the traditional process of detecting and recognizing faces, the invention can continuously identify and analyze the same person, not only can the non-cooperative face be recognizedquickly, but also the occluded and deformed face has a better recognition rate, and can be used for recognizing the non-cooperative photographed face under the video surveillance environment.

The invention relates to an energy-saving face identification device with an automatic adjustment function. The energy-saving face identification device comprises a base, a power supply plate, an identifier, a first cable, a lifting mechanism and two trigger mechanisms, the trigger mechanisms comprise trigger components, connecting components, first cylinders and first pistons, the connecting components comprise conductive boxes, conductive rods and two conductive units, the lifting mechanism comprises a motor, a connecting plate and two lifting components, and the lifting components compriseconveying belts, driven wheels, supporting units, driving shafts and driving wheels. The energy-saving face identification device with the automatic adjustment function can automatically control the identifier to be opened and closed through the trigger mechanisms according to a condition whether a user stands on a pedal or not, and the device avoids that redundant electric energy is consumed as the identifier is powered on when the user is absent, so that the device saves the energy. Compared with an existing trigger mechanism, the height of the identifier is adjusted by the lifting mechanism, a camera faces to the face of the user, a right face image is conveniently shot, and the face identification accuracy of the device is improved.

The embodiment of the invention provides a face recognition method and device, a medium and an electronic device, and belongs to the technical field of computers. The method comprises the steps of obtaining a current visible light face image and a current infrared face video of a target object; inputting the current visible light face image into a trained convolutional neural network model based on deep learning, and outputting a first recognition result; obtaining a second recognition result according to the current infrared face video; and obtaining a face recognition result of the target object according to the first recognition result and the second recognition result. According to the technical scheme provided by the embodiment of the invention, the final face recognition result can be obtained by integrating the recognition result of the visible light face image and the recognition result of the infrared face video, and the face recognition accuracy is improved.

The invention discloses a face identification based authentication method and device. One detailed description of the method includes responding to determining that a to-be-authenticated face image isin a preset face identification area, and determining attitude information corresponding to the to-be-authenticated face image; generating an attitude reference line based on the attitude information; judging whether the attitude reference line matches a preset attitude adjustment area, wherein the attitude adjustment area belongs to the face identification area; and performing face identification on the to-be-authenticated face image if yes so that whether a user corresponding to the to-be-authenticated face image is an authentication user can be determined. Thus, face identification accuracy of the to-be-authenticated face images can be enhanced.

The face recognition method comprises the steps of collecting a face image of a user; when it is detected that the face image is partially shielded, segmenting the face image according to a preset face image segmentation mode to obtain a plurality of segmented regions; performing feature comparison on an unshielded segmented region in the plurality of segmented regions and a preset segmented region of a registered user to obtain the similarity of the unshielded segmented region; and determining a face recognition result according to the similarity of the unshielded segmented regions. Therefore, the face recognition can still be effectively carried out when the face is partially shielded, and the face recognition accuracy when the face is partially shielded can be improved through the similarity comparison of the features of the non-shielded segmented regions.

The invention provides a VIP identification method based on face identification and authentication. The method comprises the following steps: step 1: a cloud server, at least one client, an image identification module and an image acquisition device are respectively arranged, wherein the cloud server performs network communication with the client, the cloud server stores VIP personnel image information, and the image acquisition device is mounted in an area needing to be monitored; step 2: the VIP personnel image information is registered into the VIP client; step 3: the image identification module extracts face features from the VIP personnel image information and stores the face features into a VIP feature library in the client; and step 4: the image acquisition device acquires the imageof the current area and transfers the acquired image into the image identification module. A VIP welcome system based on face identification and authentication combines information technology, electronic communication technology and face identification technology; and by virtue of face identification, when the VIP personnel stands erectly in front of the equipment or passes through the equipment,the system can automatically acquire the needed information.

The invention discloses a multi-template face automatic input method in a face recognition system based on a video. The method comprises the following steps: firstly, obtaining a maximum face track ina video scene, automatically storing face pictures at multiple angles as templates according to a certain screening logic, and excluding pictures with too small faces, dynamic blurring and large angles during screening, thereby ensuring the quality of face templates; secondly, conducting illumination compensation on the template which is overexposed or insufficiently exposed, adding the compensated picture into the template of the person, and improving the comparison accuracy; and finally, converting the stored template extraction features into a face feature file, storing the face feature file into a face comparison database, and directly comparing the features, thereby reducing the operation time consumption. According to the input method, automatic collection of the user face is achieved, the input time is short, the user matching degree is low, meanwhile, the input template quality is high, and the recognition accuracy can be effectively improved.

The invention belongs to the field of face recognition systems, and particularly relates to a face recognition method based on an unmanned aerial vehicle, which comprises a video stream acquisition / transmission module of an unmanned aerial vehicle end, a video stream receiving module of a ground station, a face recognition module, a face acquisition module and a database interaction module. The video stream acquisition / transmission module of the unmanned aerial vehicle end is responsible for calling the Raspberry Pi to shoot image information in a target area through a UVC camera mounted on aUSB interface; the video stream receiving module of the ground station is mainly responsible for receiving a data packet transmitted from the unmanned aerial vehicle subsystem and decoding the data packet into a format which can be recognized and processed by the human face recognition module; the face recognition module is mainly responsible for a face recognition function; and the ground stationface acquisition module provides a real-time acquisition mode and a batch import mode. The method is high in recognition precision, free of face image distortion and high in anti-interference capability.

The invention relates to a safe and reliable facial recognition ATM device. The device comprises a device body and a detecting mechanism. A display screen, a cash outlet, a card slot, a groove, a veinbox and a plurality of keys are arranged on the device body.A photographing mechanism is arranged in the groove, a collecting mechanism is arranged in the vein box, the collecting mechanism comprisesa sensing assembly and a collecting assembly, the collecting assembly comprises an image sensor, a protective hood and two emission units, the emission units comprise lamp tubes, lamp hoods, direction adjusting rods and direction adjusting units, and the photographing mechanism comprises a camera, a sleeve tube, a framework, a driving assembly and two adjusting assemblies. According to the ATM device, collecting imaging is conducted on a palm vein through the collecting mechanism, vivo detection is achieved to prevent other people from personating facial images of a user to operate an ATM, the properties of the user are protected, moreover, the angle of the sleeve tube is adjusted through the photographing mechanism, and the camera faces directlythe face of the user to collect images fromthe front side to improve the accuracy of facial recognition and the practicability of equipment.

The invention discloses a model training method, a face recognition method and device, equipment and a medium, and relates to the technical field of artificial intelligence such as computer vision anddeep learning. According to the specific implementation scheme, the method comprises the steps of: obtaining at least two training data sets, wherein each training data set comprises face sample datawith the same number, and the types of the face sample data in different training data sets are different; performing face prediction on each training data set by using a pre-established face recognition model to obtain a face prediction result of each training data set; and performing supervised training on the face recognition model in parallel for the face prediction result of each training data set by utilizing pre-labeled face labeling data of each training data set. The face recognition model obtained by training can improve the face recognition accuracy of the sample data with less data distribution.

The invention discloses a four-value weight and multiple classification-based human face feature extraction method. The method comprises the steps of constructing a human face training sample database; establishing a convolutional neural network; adjusting a caffe framework; preprocessing a human face image sample, inputting the human face image sample to the convolutional neural network for performing training, until the network is completely converged, and storing a generated human face identification model; and preprocessing a to-be-extracted human face image, producing a mean value file, inputting the mean value file to the human face identification model to obtain a feature graph, rotating the feature graph for multiple different angles to extract features respectively, performing addition fusion on different angle features of the same image, and finally obtaining a main human face feature. The method has the beneficial effects that the problems of huge memory consumption and insufficient storage space of network training is radically solved; and the feature with a strongest expression capability is obtained in a multi-feature extraction fusion mode, so that the human face identification accuracy is remarkably improved.

The embodiment of the invention discloses a neural network processing and face recognition method and device, equipment and a storage medium, and the method comprises the steps: obtaining a first neural network and a face image, and enabling the first neural network to represent a neural network for face recognition; obtaining annotation information in the face image; performing incremental training on the basis of the first neural network according to the optimization data to obtain a second neural network, wherein the optimization data at least comprises the face image with the annotation information. Thus, in the embodiment of the invention, incremental training can be automatically carried out based on the face image with the annotation information on the basis of the first neural network, so that the neural network capable of recognizing the face image is obtained, and the face recognition requirement of the face image in a real environment can be met.

The invention discloses a face recognition method based on dictionary learning models. The method comprises the following steps of: mapping trained and tested face images to a low-dimension space to acquire a training signal set matrix; establishing the dictionary learning models which comprise an irrelevant dictionary learning (IDL) model and an unconstrained irrelevant dictionary learning (U-IDL) model; inputting the training signal set matrix into the IDL and U-IDL models, and solving the models to acquire an irrelevant dictionary and a linear classifier; acquiring a corresponding sparse vector of each picture belonging to a test sample based on the dictionary acquired in the last step by using a sparse expression algorithm; and inputting the sparse vectors into the linear classifier to acquire category labels of test sample pictures, wherein the result expressed by the category labels is used as the face recognition result. The invention provides the new models and the new method for dictionary learning problems in sparse expression, and the models and the method can be applied to mode identification and image classification problem under common conditions; and particularly, aiming at face recognition application, the dictionary learning method can achieve relatively high face recognition accuracy.

The invention relates to a face recognition device convenient to adjust and used for vehicle access. The device comprises a mounting plate, a contraction mechanism, a connecting rod, a recognition plate, a recognition mechanism and two mounting components. The contraction mechanism comprises a driving component, a first bevel gear, a second bevel gear, a rotating rod, a connecting frame and two insertion tubes; the driving component comprises a translation unit, a first rack, a circular gear, a rotating shaft and a supporting unit; the recognition mechanism comprises a camera, a moving component, a moving plate and a direction adjusting component; and the moving component comprises a first motor, a frame and two moving units. According to the face recognition device convenient to adjust and used for vehicle access, the position of the recognition plate can be flexibly adjusted through the contraction mechanism according to the requirement, thus face recognition is convenient and the wide view field of a vehicle owner during driving is ensured, the recognition mechanism drives the camera to move and swing, thus a front face image of the vehicle owner is conveniently acquired, and the face recognition precision and the practicability of the equipment are improved.

The invention discloses a face recognition system based on silent living body detection. The face recognition system comprises a data acquisition module, a data preprocessing module, a face detection module, a living body detection module and a face recognition module; the living body detection module carries out living body detection by using models of three levels, respectively carries out living body or attack classification on the whole photo, carries out living body or attack classification on the face area photo and carries out attack feature detection on the whole photo; the detected objects comprise moire patterns, abnormal light and shadow, mobile phone frames, photograph holding, display light reflection, plastic material light reflection and the like; according to the face recognition system, the attack or the living body can be distinguished only by performing transient shooting on the front face of the user towards the screen, the model in the living body detection module has good generalization ability and high recognition precision, the attack success probability is effectively reduced, the user safety is ensured, and the invention is suitable for various face recognition scenes such as identity authentication and unlocking login.

The invention discloses a face recognition method and device, equipment and a storage medium, belonging to the technical field of image processing, in particular to the technical field of computer vision. The face recognition method comprises the steps of acquiring a to-be-recognized face image, wherein glasses are worn on the face in the to-be-recognized face image; determining a spatial relativeposition of a glasses frame and the face in the to-be-recognized face image and a wearing inclination angle of the glasses; and performing face recognition according to the spatial relative positionsof the glasses frame and the face, the wearing inclination angle of the glasses and the feature points. According to embodiments of the invention, the face recognition accuracy in a scene that a human face wears glasses can be improved.

The invention provides a face recognition device having anti-interference capability and guaranteeing comprehensive feature extraction. The face recognition device has a body, a pedal and an enclosuremechanism. The enclosure mechanism comprises a top plate, gauze, a plurality of lamp tubes, two supporting mechanisms and four baffle plates. Each supporting mechanism comprises a side plate, a firstmotor and two supporting assemblies; each supporting assembly comprises a driving unit, a telescopic rack, a hinge block and two connecting rods. A shooting mechanism comprises a shooting plate, a second motor, two lifting assemblies and a plurality of second cameras; each lifting assembly comprises a second driving shaft, a driving wheel, a pull wire, a lifting block and a guide unit. The face recognition device having anti-interference capability and guaranteeing comprehensive feature extraction has the following advantages: during identity authentication, an enclosed environment is formedvia the enclosure mechanism, so that the device is prevented from being influenced by external light during a face recognition process; the shooting mechanism can be used for compensation shooting forareas around a nose, so that complete face feature data can be extracted by the apparatus and the face recognition precision of the apparatus is further improved.

The present invention discloses a screen unlocking method, apparatus and device, and a computer readable storage medium. The method comprises the steps of: employing an atomization detector under outside triggering to detect whether a lens of a depth camera device is atomized or not, if yes, starting a heating device to heat the lens, or else, employing the depth camera device to collect 3D information of a face in a corresponding area; and after the 3D information of the face in the corresponding area is collected, determining whether pre-stored 3D information has information being consistentto the collected 3D information or not, if yes, performing unlocking operation of a screen of a corresponding terminal, or else, outputting prompting information that the collected 3D information hasno unlocking authority. The face recognition precision is greatly increased when the screen is unlocked. Besides, influences of human factors and weather factor are reduced through determination of atomization, the determination accuracy is improved, and the working efficiency is improved.

The invention relates to a face recognition ATM with a good recognition effect, and the ATM comprises a housing, a shading mechanism and two cameras. The interior of the housing is provided with an adjusting mechanism and two sliding rails. The shading mechanism comprises a lifting assembly, a reciprocating component and two shading plates. The lifting assembly comprises a positioning rod, a driving unit, a telescopic frame, a connecting plate, two sliding blocks, two supporting plates and two transmission blocks. The adjusting mechanism comprises a first motor and two adjusting assemblies, and each adjusting assembly comprises a first gear, a rotating disc and a connecting rod. The driving unit comprises a second motor, a second gear and two transmission units. The ATM achieves the shading of strong light through the shading mechanism, and reduces the impact on the face recognition of the ATM from strong light, thereby improving the recognition effect of the ATM. The shooting angles of the cameras are adjusted through the adjusting mechanism to collect the face data at different angles for comparison, so that the face recognition precision of the ATM is higher.

The invention relates to an easy-to-adjust vehicle access device based on face recognition, which comprises a mounting plate, an adjusting mechanism, a main body and four suction cups, wherein a camera and two lamp shells are arranged on the main body; a PLC and a power supply mechanism are arranged in the main body; the adjusting mechanism comprises a steering assembly, a strut, a hinge unit, a connecting rod, an air pump, a cylinder and a piston; the power supply mechanism comprises a battery, a circuit breaker, a sliding wire rheostat and a translating assembly; the translating assembly comprises a first motor, a gear and a rack; the easy-to-adjust vehicle access device based on face recognition can control the angle position of the main body through the adjusting mechanism; the main body rotates upward when driving to avoid obscuring the line of sight of the driver; after being parked, the main body rotates downward, so that the camera can collect the front image for face recognition. Moreover, the voltage at both ends of the lamp tube is adjusted through the power supply mechanism, so that the camera photographs under stable illumination, thereby improving the accuracy of facerecognition and the practicability of the device.

The present invention discloses a face recognition system of a robot based on a time frame. The system comprises: a collection unit configured to collect a face image of a person to be recognized; anextraction unit configured to extract a face contour according to the face image; a time unit configured to obtain a current moment of extracting the face contour; a control unit configured to determine whether the face contour meets a preset condition or not and output corresponding control signals according to a determination result; a processing unit configured to enter a starting state or a standby state according to the control signals sent by the control unit, wherein when the processing unit is in the starting state, a recognition unit is configured to perform recognition of the face contour according to a standard face contour at a time frame corresponding to the current moment and output a recognition result; and an interaction unit configured to execute corresponding man-machineinteraction operation according to the recognition result.

The invention discloses a face correction method and system of a generative adversarial network based on identity constraints. The generative adversarial network based on the identity constraints is constructed and trained. The network comprises a generator, a discriminator and a face feature extraction network. The generator comprises an encoder, an angle posture classification module, an identity constraint recognition module and a decoder. By designing an angle posture classification module and an identity constraint recognition module,the posture coding features and identity coding features are respectively acquired, and a feature loss function is introduced to constrain the identity coding features of the side face to approach the identity coding features of the front face; then identity constraints are added to the side face coding features, and decoupling of identity information and posture information in the side face coding features is realized, so that the side face identity coding features can generate a front face image with identity consistency through a decoder.

The invention relates to an anti-interference high-precision face recognition device. The device comprises a shell, a pressing plate and a sealing mechanism, the sealing mechanism comprises a ceiling,a gauze, a plurality of illuminating lamps, two supporting mechanisms and four shielding plates, and the supporting mechanism comprises a side plate, a first motor and two supporting assemblies. Thesupporting assembly comprises a driving unit, a telescopic frame, a connecting block and two connecting rods, a shooting mechanism comprises a flat plate, a second motor, two lifting assemblies and aplurality of second cameras, and each lifting assembly comprises a rotating shaft, a rotating wheel, a connecting wire, a lifting block and a guide unit. According to the anti-interference high-precision face recognition device, a closed environment is formed through a closed mechanism during identity authentication, so that the device is prevented from being interfered by the external light during the face recognition process. In addition, the compensation shooting is carried out at the periphery of a nose through the shooting mechanism, the device can extract the complete face feature data conveniently, and the face recognition precision of the device is further improved.

The invention provides a face recognition method and system based on a block chain, and belongs to the technical field of block chains. The face recognition method based on the block chain comprises the steps of obtaining current face image data of a user; inputting the current face image data into an optimal face recognition model to obtain a current face recognition result; wherein the step of creating the optimal face recognition model comprises the following sub-steps: uploading face recognition model parameters to a block chain network to enable the block chain network to generate globalmodel parameters according to the face recognition model parameters; and creating an optimal face recognition model according to the global model parameters. According to the invention, the privacy data of the user can be protected while the face recognition accuracy is improved.

The invention discloses a face recognition method under a low resolution condition. M low-resolution facial image samples and a to-be-recognized facial image are normalized to a preset size, sliding block LBP feature vectors are respectively extracted, and an extraction method comprises the following steps: getting LBP vector images of the facial images, obtaining an LBP histogram feature vector of each sliding block area by adopting sliding block transverse and statistics, arranging the LBP histogram feature vectors according to serial numbers of the sliding block areas, and obtaining the sliding block LBP feature vectors of the facial images; and the sliding block LBP feature vectors of the M facial image samples and the sliding block LBP feature vector of the to-be-recognized facial image are constructed into a matrix, then dimensionality reduction processing is performed, and recognition is performed according to the feature vectors in a feature matrix after dimensionality reduction. The face recognition method disclosed by the invention can fully extract features of low-resolution facial images, and face recognition accuracy rate under the low resolution condition is improved.

The invention discloses a human face super-resolution reconstruction recognition method based on fractional order orthogonal partial least squares. The method comprises the following steps: 1, readjusting intra-group and inter-group covariance matrixes through fractional order eigenvalues and singular values by using fractional order orthogonal partial least squares, calculating a projection direction, mapping face image features into subspaces, and reconstructing low-resolution input high-resolution global face features through a neighborhood reconstruction idea; 2, constructing a high-resolution face residual block by using a neighborhood reconstruction method, synthesizing the residual block to obtain high-resolution face residual compensation, and supplementing face details through a residual compensation strategy; and 3, adding residual compensation to the global face to form a high-resolution face image output by the final algorithm, wherein the high-resolution global face features can be used for face recognition. In the face super-resolution reconstruction and recognition application, a better face reconstruction effect and higher face recognition accuracy can be obtained.